Bugfix/sorting97 (#551)

* Prevent sorting for list of input formats * Update subversion
Ensembl · Aug 6, 2019 · d896ba4 · d896ba4
1 parent 8bf4349
commit d896ba4
Show file tree

Hide file tree

Showing 4 changed files with 26 additions and 3 deletions.
diff --git a/modules/Bio/EnsEMBL/VEP/Config.pm b/modules/Bio/EnsEMBL/VEP/Config.pm
@@ -396,6 +396,12 @@ our %DEPRECATED = (
   'convert' => undef,
 );
 
+our %UNSORTABLE = (
+  id     => 1,
+  hgvs   => 1,
+  spdi   => 1,
+  region => 1,
+);
 
 ####################################
 ####################################
@@ -470,6 +476,8 @@ sub new {
     $config->{$flag} = [split(',', $config->{$flag})];
   }
 
+  $config->{unsorted_formats} = \%UNSORTABLE;
+
   $self->apply_option_sets($config);
 
   # check config before we return

diff --git a/modules/Bio/EnsEMBL/VEP/Constants.pm b/modules/Bio/EnsEMBL/VEP/Constants.pm
@@ -53,7 +53,7 @@ use warnings;
 use base qw(Exporter);
 
 our $VEP_VERSION     = 97;
-our $VEP_SUB_VERSION = 2;
+our $VEP_SUB_VERSION = 3;
 
 our @EXPORT_OK = qw(
   @FLAG_FIELDS

diff --git a/modules/Bio/EnsEMBL/VEP/InputBuffer.pm b/modules/Bio/EnsEMBL/VEP/InputBuffer.pm
@@ -203,7 +203,9 @@ sub next {
       $prev_chr = $vf->{chr};
     }
   }
-
+
+  my $unsorted_formats = $self->config->{_params}->{unsorted_formats};
+
   if(my $parser = $self->parser) {
     while(@$buffer < $buffer_size && (my $vf = $parser->next)) {
 
@@ -234,7 +236,7 @@ sub next {
       else {
         push @$buffer, $vf;
         $prev_chr = $vf->{chr};
-        if (!$self->param('no_check_variants_order') && $self->param('format') ne 'id') {
+        if (!$self->param('no_check_variants_order') && !$unsorted_formats->{$self->param('format')}) {
           # Use a default distance to check if the variant is still in the same region
           # even if it's not ordered with the previous variant location: we use the VF start and not the VCF start
           # (they can be different when the variant is a deletion for instance and/or when the alleles can be minimised)

diff --git a/t/InputBuffer.t b/t/InputBuffer.t
@@ -502,6 +502,19 @@ eval {
 };
 ok(!$@, "Skip the count for non ordered variants with the flag 'no_check_variants_order'");
 
+# Skip check for hgvs variants
+# To workaround the offline check for HGVS input, the parser being given to the InputBuffer is in VCF format. The sorting is correctly prevented by the InputBuffer, compared to the previous 2 tests.
+$cfg = Bio::EnsEMBL::VEP::Config->new({%{$test_cfg->base_testing_cfg}, buffer_size => 100, format => 'hgvs'});
+$p = Bio::EnsEMBL::VEP::Parser::VCF->new({config => $cfg, file => $test_cfg->{not_ord_vcf}, valid_chromosomes => [1,21,22]});
+$ib = Bio::EnsEMBL::VEP::InputBuffer->new({config => $cfg, parser => $p, max_not_ordered_variants => $max_non_ordered_variants, max_not_ordered_variants_distance => $max_not_ordered_variants_distance});
+
+$ib->next();
+$ib->next();
+eval {
+  $ib->next();
+};
+ok(!$@, "Skip the count for non ordered variants with the flag 'hgvs'");
+
 
 # Check for non ordered variants with larger position difference (default settings)
 $max_not_ordered_variants_distance = 100;